I have come across an essay title asking us to critique the evidence of language being processed as either “abstract phonemes” or “surface exemplars”. (Specifically in phonology) Is this a rewording of the debate between OT vs RBT? Rules vs. constraints? Generativism/Chomsky vs. ? ? I feel lost with what to plan.

Could anyone clarify the actual “standard” names of the relevant theories or point towards any particular papers?

Would anyone be able to outline the two opposing sides? There seem to be so many different terms thrown around that it is difficult to split everything into two opposing sides.

Thanks for any input you have, as I have been searching for literally weeks.

  • 1
    The word phoneme himself means an abstract or cognitive reality in comparison to phonetic which is the study of the physical reality of the language. So talking about "abstract phoneme" is a pleonasm.
    – amegnunsen
    Commented Jan 4, 2019 at 15:15
  • 1
    I cannot articulate how much that elucidates the question! Thank you Commented Jan 4, 2019 at 16:38
  • (Any further comments would be much appreciated) Commented Jan 4, 2019 at 16:39
  • It would seem that "processed" is the key word. Children learning to speak are confronted not with abstract phonemes but with surface exemplars of the language they eventually come to speak. After they become speakers, they presumably have an abstract phoneme system that allows them to make up exemplars appropriately as needed.
    – jlawler
    Commented Jan 4, 2019 at 17:26
  • That is an interesting point. It seems almost as though the “exemplar theory” side will have to argue that children store new words’ pronunciation without any “compression” – which in itself sounds to be a challenging stance to defend. Commented Jan 4, 2019 at 17:37

2 Answers 2


I hear tell that young children sometimes get mad at adults who talk to them in child speech, because they can tell they're being made fun of. Adult says "And do we hear the fwoggy cwoaking?", and child betrays signs of irritation: "No, it's not fwoggy. It's fwoggy!"

If this happens, it suggests that we don't even hear surface exemplars, usually.

  • Fair point you make. I came across a similar set of examples in “Words in the Mind” by Jean Aitchison but had never made this connection. Perhaps it could also spark discussion on perceptive phonemic boundaries (e.g. do we recognize the [w] in “fwoggy” as an allophone of the /r/ as in “froggy”, or is it perceived as within the boundary of the one phone [ɹ] and therefore as the same sound? Unless both of the aforementioned qualify as “allophone” phenomena...) Commented Jan 4, 2019 at 18:19
  • @tiredstudent I would say, no. We recognize it as disordered speech. We recognize that the speaker uses the wrong phoneme. In this case even the child seems to recognize that but it hasn't mastered the articulation (the perception is ahead). Or the child still makes some distinction that we don't perceive and the child's phone is an allophone of /ɹ/ for the child but an allophone of /w/ for us whereas the adult's [w] is an allophone of /w/ for both the adult and the child. Commented Jan 4, 2019 at 22:38

All of the terminology that you propose is "the same" in fundamental ways that make those theories distinct from Exemplar Theory.

The hard part for you will be figuring out what "exemplar theory" says, as applied to phonology. Outside of exemplar theoretic approaches (there are many of them), everybody has "abstract phonemes", that is some kind of mental representation of sound. In fact, an exemplar is also a mental representation (the only thing that isn't abstract is an actual waveform, and you don't have them in your head); it's just claimed to be less abstract.

Indeed, it is not clear that exemplar approaches are "phonology" in the same sense that OT or rule-based theory is, that is, it may not be a theory of how to compute Arabic verb inflection (Arabic has an interestingly complex set of alternations, often called "morphophonology" because it is phonology connected to morphological differences such as the different effects of adding the endings /t,na/ versus /at,aw/ to /bana/: [baneet, baneena] vs. [banat, banaw]). Exemplar theory seems to address the question "how are you likely to actually pronounce this word", rather that "what is the form of the 3rd pl perfective of 'build'?". That is, EP and generic computational theories of phonology with phonemes are about different things. EP focuses on things like child language acquisition, accommodation to dialect and individual differences and production variation. Grammatical theories may indeed embrace those sorts of phenomena. Frequently, people point to some fact of child language or slips of the tongue to support their claimed generalization. Some people e.g. Hale and Reiss and numerous others distinguish between facts about grammar – phonology – versus speech and performance – phonetics, maybe.

Probably the best reference for applying exemplar theory to phonology is this paper by Johnson.

  • Fascinating ideas; thank you very much for your reply. The crossover into morphology is interesting and indeed, as you, say a challenge when instructed to remain within the phonological realm. Commented Jan 4, 2019 at 17:41
  • I wonder whether you or anyone else would be able to point me in the direction of some exemplar theory papers/literature/authors, for finding some of the proposed evidence? Commented Jan 4, 2019 at 17:44
  • It seems that "surface" is simply a synonym of morphology, it is not opposed necessarily to phoneme.
    – amegnunsen
    Commented Jan 4, 2019 at 18:37
  • I'm trying to make sense of the Johnson paper you referred us to. No luck, so far.
    – Greg Lee
    Commented Jan 4, 2019 at 20:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.